Mechanics-Guided Embryonic Patterning of Neuroectoderm Tissue from Human Pluripotent Stem Cells

Home , Neural crest, Neural plate, Neuroectoderm, Neuroepithelial cell, Neurulation

Articles https://doi.org/10.1038/s41563-018-0082-9

Mechanics-guided embryonic patterning of neuroectoderm tissue from human pluripotent stem cells

Xufeng Xue1,9, Yubing Sun1,2,9*, Agnes M. Resto-Irizarry1, Ye Yuan3, Koh Meng Aw Yong1, Yi Zheng1, Shinuo Weng 1, Yue Shao 1, Yimin Chai4, Lorenz Studer5,6 and Jianping Fu 1,7,8*

Classic embryological studies have successfully applied genetics and cell biology principles to understand embryonic development. However, it remains unresolved how mechanics, as an integral driver of development, is involved in controlling tissue-scale cell fate patterning. Here we report a micropatterned human pluripotent stem (hPS)-cell-based neuroectoderm developmental model, in which pre-patterned geometrical confinement induces emergent patterning of neuroepithelial and neural plate border cells, mimicking neuroectoderm regionalization during early neurulation in vivo. In this hPS-cell-based neuroectoderm patterning model, two tissue-scale morphogenetic signals—cell shape and cytoskeletal contractile force—instruct neuroepithelial/ neural plate border patterning via BMP-SMAD signalling. We further show that ectopic mechanical activation and exogenous BMP signalling modulation are sufficient to perturb neuroepithelial/neural plate border patterning. This study provides a useful microengineered, hPS-cell-based model with which to understand the biomechanical principles that guide neuroectoderm patterning and hence to study neural development and disease.

ne of the enduring mysteries of biology is tissue morpho- on glass coverslips (Fig. 1a and Supplementary Fig. 1). H1 human genesis and patterning, where embryonic cells act in a embryonic stem (hES) cells were plated as single cells at 20,000 coordinated fashion to shape the body plan of multicellu- cells cm−2 on adhesive islands to establish micropatterned colonies O 1–5 lar animals . As a highly conserved developmental event crucial with a defined circular shape and size. A differentiation medium for the nervous system formation, neural induction, for example, supplemented with the dual SMAD inhibitors, SB 431542 (SB, leads to differentiation of the ectoderm into a patterned tissue, con- TGF-β inhibitor; 10 µM) and LDN 193189 (LDN, BMP4 inhibitor; taining the neuroectoderm (neural plate, or NP) and the epidermal 500 nM), was applied for neural induction22 (Supplementary Fig. 1; ectoderm separated by the neural plate border (NPB) (Fig. 1a)6,7. see Methods). The β-catenin stabilizer CHIR 99021 (CHIR, 3 µM), Classic embryological studies of neural induction have unravelled a WNT activator, was also supplemented in culture (Supplementary the importance of graded developmental signalling mediated by Fig. 1). CHIR promotes NPB cell specification under the neural diffusible signals such as bone morphogenetic proteins (BMPs) induction condition established by the dual SMAD inhibitors23,24. (Fig. 1a)8–10. However, neural induction, like any tissue-scale mor- Although cells distributed uniformly on adhesive islands 24 h phogenetic event, occurs within the milieu of biophysical determi- after initial cell plating, neural induction resulted in differentiat- nants, including changes in shape, number, position and force of ing cells gradually accumulating in the colony central area, lead- cells7,11. Yet it remains undetermined how these tissue-scale mor- ing to a significantly greater cell density at the colony centre than phogenetic changes work in concert with classic developmental at the periphery (Fig. 1b and Supplementary Fig. 1). Cell density signaling events mediated by diffusible signals for proper cell fate was further analysed using DAPI fluorescence intensity. The full patterning during neural induction. width at half maximum (FWHM) for spatial distributions of DAPI hPS cells, which reside in a developmental state similar to the intensity decreased continuously from 336 µm at day 1 to 240 µm pluripotent epiblast12,13, have been successfully used for the devel- at day 9 (Supplementary Fig. 1). Confocal images further showed opment of self-organized models of human embryonic develop- that micropatterned colonies at day 7 remained as a monolayer. ment14–21. To date, however, no neural induction models exist Strikingly, quantitation of colony thickness and nucleus shape that leverage hPS cells and their innate self-organizing properties revealed that at this point, cells exhibited a gradual change of cell to study neuroectoderm patterning. Here, we sought to develop shape from a columnar phenotype with columnar nuclei at the col- micropatterned hPS-cell colonies on two-dimensional substrates to ony centre to a cuboidal morphology with rounded nuclei at the model neural induction. Microcontact printing was utilized to gen- colony periphery (Supplementary Fig. 1), consistent with character- erate vitronectin-coated, circular adhesive islands with a diameter istic neuroectoderm thickening during neural induction in vivo6,11. of 400 µm on flat poly-dimethylsiloxane (PDMS) surfaces coated Importantly, Pax6+ neuroepithelial (NE) cells, the neural progenitor

1Department of Mechanical Engineering, University of Michigan, Ann Arbor, MI, USA. 2Department of Mechanical and Industrial Engineering, University of Massachusetts, Amherst, MA, USA. 3School of the Gifted Young, University of Science and Technology of China, Hefei, China. 4Department of Orthopedic Surgery, Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, Shanghai, China. 5Developmental Biology Program, Memorial Sloan- Kettering Institute, New York, NY, USA. 6Center of Stem Cell Biology, Memorial Sloan-Kettering Institute, New York, NY, USA. 7Department of Biomedical Engineering, University of Michigan, Ann Arbor, MI, USA. 8Department of Cell and Developmental Biology, University of Michigan Medical School, Ann Arbor, MI, USA. 9These authors contributed equally: Xufeng Xue, Yubing Sun *e-mail: [email protected]; [email protected]

Nature Materials | VOL 17 | JULY 2018 | 633–641 | www.nature.com/naturematerials 633 © 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved. Articles NATure MATerIAlS a b Day: 136 9

In vivo In vitro

hPS cells Phase Ectoderm Mesoderm PDMS Endoderm Cover slip Neural induction Neural plate (neuroectoderm) Neural plate DAPI NPB NPB Epidermal ectoderm BMP activity Micropatterned High Low High cell colony Intensity map ncolony = 24 ncolony = 35 ncolony = 20 ncolony = 59

0–1 (arbitrary units) c PAX6 PAX3 ZIC1 MSX1 d Day: 4 678 Phase Staining image GFP Intensity map Intensity ma p ncolony = 132 ncolony = 30 ncolony = 49 ncolony = 53 ncolony = 60 ncolony = 60 ncolony = 60 ncolony = 60 Relative intensity ma p 0–1 (arbitrary units) 0–1 (arbitrary units)

Fig. 1 | Self-organized neuroectoderm patterning in circular hPS cell colonies. a, Schematic of neural induction in vivo and in vitro. During neural induction in vivo, embryonic cells in the ectoderm form the neural plate (neuroectoderm). Embryonic cells at the neural plate border (NPB) separate the neuroectoderm from the epidermal ectoderm. Neural induction of circular hPS cell colonies leads to autonomously patterned neuroectoderm tissues, with neural plate cells in the colony central region and NPB cells at the colony periphery. b, Representative phase contrast and fluorescence images showing cell morphology and nuclei (stained by DAPI), respectively, on different days. White dashed lines mark the colony periphery. Experiments were repeated three times with similar results. Bottom average intensity maps show spatial distributions of DAPI intensity. The number of colonies analysed was pooled from n = 3 independent experiments. Data plotted are mean values. c, Representative immunofluorescence micrographs and average intensity maps showing colonies at day 9 stained for neuroectoderm marker Pax6 and NPB markers Pax3, Zic1 and Msx1. White dashed lines mark the colony periphery. Experiments were repeated three times with similar results. Relative intensity maps were normalized to DAPI signals. The number of colonies analysed was pooled from n = 3 independent experiments. Data plotted are mean values. d, Representative phase contrast and fluorescence images and average intensity maps from live cell assays using Sox10-EGFP hES cells. White dashed lines mark the colony periphery. White arrowheads mark GFP+ cells at the colony border. Experiments were repeated three times with similar results. The number of colonies analysed was pooled from n = 3 independent experiments. Data plotted are mean values. Scale bars in b–d, 100 µm.

in the NP, were found preferentially localized at the colony central tion is consistent with in vivo findings that ZIC1+ cells also exist in region on day 9, whereas Pax3+, Zic1+ and Msx1+ NPB cells were con- the neurogenic placode region25. Immunostaining and immunob- centrated at the colony periphery, forming a concentric ring-shaped lotting of E-cadherin and N-cadherin at day 2 and day 7 further tissue sheet consistent with neuroectoderm patterning, with proper confirmed successful neural conversion (Supplementary Fig. 2). To regionalization of NE and NPB cells (Fig. 1c and Supplementary confirm differentiation potentials of neuroectoderm tissues after Fig. 1). Interestingly, the concentric zone of Pax3+ cells at the neural induction, they were continuously cultured under a motor colony periphery exhibited a smaller zone width, as character- neuron differentiation medium26 or a neural crest differentiation ized by FWHM for spatial distribution of Pax3 intensity (92 µm), medium23,24 (Supplementary Fig. 3). Importantly, only putative NE compared with concentric zones of Zic1+ (128 µm) and Msx1+ cells at the colony central region differentiated into Olig2+ motor (175 µm) cells, respectively (Supplementary Fig. 1). This observa- neuron progenitor cells, whereas only putative NPB cells at the

634 Nature Materials | VOL 17 | JULY 2018 | 633–641 | www.nature.com/naturematerials © 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved. NATure MATerIAlS Articles colony periphery differentiated into Ap2α+ and Sox10+ neural crest Gradual cell accumulation at the colony centre (Fig. 1b) sug- cells (Supplementary Fig. 3). gests cell rearrangement and dynamic morphogenetic processes. We next sought to examine the effect of colony size on self- Indeed, colonies stained for ZO-1 for visualization of cell–cell junc- organized neuroectoderm patterning. To this end, circular adhesive tions revealed that at day 4, cells at the colony periphery displayed islands with diameters of 300, 400, 500 and 800 µm were fabricated a greater projected area than at the colony centre (Supplementary and compared. Neural induction resulted in the emergence of con- Fig. 9, Supplementary Fig. 10, and Fig. 2a,b). Compared with those centric, ring-shaped neuroectoderm tissues with proper regional- at the colony centre, cells at the colony periphery also displayed ization of NE and NPB cells at day 9 for all colonies (Supplementary greater intracellular cytoskeletal contractility, as revealed by stron- Fig. 4). Interestingly, the Pax6+ NE circular pattern size and the con- ger staining for phosphorylated myosin and greater traction stress centric zone width of Pax3+ NPB cells, as characterized by FWHM quantitated using PDMS micropost force sensors30 (Fig. 2a,c and for spatial distributions of Pax6 and Pax3 intensities, respectively, Supplementary Fig. 9; see Methods). Supplementing neural induc- appeared relatively constant for colony diameters of 300–500 µm tion medium with blebbistatin (10 µM), which inhibits myosin (Supplementary Fig. 4). Notably, neural induction for colonies with motor activity and thus cytoskeletal contractility, effectively elimi- a diameter of 800 µm led to marked cell accumulation at the colony nated spatial differences in projected area or traction stress between periphery on day 9, coinciding with a notable number of Pax3+, Zic1+ colony peripheral and central regions at day 4 (Supplementary and Msx1+ NPB cells at the colony central region (Supplementary Fig. 10 and Fig. 2a–c). We note that blebbistatin treatment did not Fig. 4). The underlying mechanism leading to this different cell result in notable cell toxicity (Supplementary Fig. 11). However, it fate distribution remains unclear, and probably involves second- decreased cell proliferation and thus effectively increased the cell ary tissue morphogenesis18. We further examined neuroectoderm spreading area at the colony central area (Supplementary Fig. 11). patterning in circular colonies of 400 µm in diameter under two Importantly, even though NE and NBP cells still appeared at day other initial cell seeding density conditions (5,000 cells cm−2 and 9 with blebbistatin treatment, proper neuroectoderm regionaliza- 30,000 cells cm−2). A seeding density of 5,000 cells cm−2 led to abnor- tion was severely impaired, with abundant Pax3+, Zic1+ and Msx1+ mal neuroectoderm regionalization. Even though a notable num- NPB cells randomly distributed across entire colonies (Fig. 2d). ber of Pax6+ NE cells appeared at the colony centre, Pax3+, Zic1+ Reducing cell seeding density under default neural induction con- and Msx1+ NPB cells were evident across entire colonies without ditions to 5,000 cells cm−2 also caused more uniform spatial distri- distinct spatial patterning (Supplementary Fig. 5). Plating cells at butions of projected area and traction stress across entire colonies 30,000 cells cm−2 led to the development of multilayered cellular (Supplementary Fig. 10). structures by day 9, even though Pax3+ NPB cells still appeared and Our results in Fig. 2 suggested critical roles of morphogenetic formed a single peripheral layer enveloping the colony top surface cues in mediating neuroectoderm patterning with proper region- (Supplementary Fig. 5). Together, we identified micropatterned cir- alization. Neural induction in vivo has been shown to be mediated cular colonies of 300–500 µm in diameter and an initial cell seed- through graded BMP signalling8–10,31. Specifically, a BMP signalling ing density of 20,000 cells cm−2 as the most suitable conditions for gradient provides positional information in the ectoderm, with high in vitro modelling of neural induction (referred to henceforth as BMP activity promoting epidermal differentiation, low BMP for NP the ‘default neural induction’ condition). Emergent patterning development, and intermediate BMP for NPB specification8–10,31 of neuroectoderm tissues with proper autonomous regionaliza- (Fig. 1a). Indeed, we observed strong correlations between spatial tion of NE and NPB cells was also achieved using another hES cell regulations of cell shape, cytoskeletal contractility and BMP activ- line (H9) and a human induced pluripotent stem (hiPS) cell line ity at day 4. Most cells at the colony periphery showed prominent (Supplementary Fig. 6). nuclear staining of phosphorylated SMAD 1/5 (p-SMAD 1/5), a We further conducted live-cell assays to examine dynamic neu- downstream target of BMP-SMAD signalling, whereas far fewer roectoderm patterning using a Sox10-EGFP bacterial artificial cells at the colony central area were p-SMAD 1/5 nuclear positive chromosome hES cell reporter line (H9)23. SOX10 is a specifier gene (Fig. 3a,b). Interestingly, blebbistatin treatment or plating cells at for neural crest induction at the NPB in vertebrates27. Surprisingly, 5,000 cells cm−2 rescued BMP activity at the colony central area on under the default neural induction condition, GFP+ cells emerged day 4 (Fig. 3a,b), consistent with their effects on promoting NPB first at the colony periphery on day 6 and continuously increased in differentiation at the colony central region. Replacing LDN in the number there through day 7 and day 8 (Fig. 1d). This observation default neural induction condition with NOGGIN (100 ng ml−1), was further corroborated by immunofluorescence analysis to exam- a protein that antagonizes BMPs, or dorsomorphin (1 µM), which ine initial appearances and spatial distributions of Pax6+ NE and inhibits BMP receptors, led to similar results of graded SMAD tran- Pax3+ NPB cells (Supplementary Fig. 7). Together, the time course scriptional activation at day 4 and neuroectoderm regionalization at of expression of NPB markers Pax3, Zic1 and Msx1 and neural crest day 9 (Supplementary Fig. 12). marker Sox10 in micropatterned hES cell colonies is consistent with To specifically examine the functional roles of cell shape and cyto- mouse embryo development in vivo27–29. skeletal contractility in BMP activation, microcontact printing was To examine a possible role of cell sorting in neuroectoderm pat- applied to obtain patterned circular single hES cells with prescribed terning, live-cell imaging was conducted for tracking cell migratory spreading areas (Supplementary Fig. 13 and Fig. 3c). Cytoskeletal behaviours. Migration of cells, whether located at the colony centre contractility of patterned single hES cells correlated positively with or at peripheral region, was very limited under the default neural spreading area (Supplementary Fig. 13). Importantly, the percent- induction condition, with an average radial displacement less than age of patterned single hES cells with dominant nuclear staining 20 µm between days 2 and 4 (Supplementary Fig. 8). Furthermore, of p-SMAD 1/5 also increased with spreading area (Fig. 3c,d). cell migration could be well simulated using an unbiased random Furthermore, quantitative polymerase chain reaction with reverse walk model (Supplementary Fig. 8; see Methods). Thus, cell sorting transcription (qRT-PCR) analysis revealed greater expression levels was unlikely to be responsible for emergent regional neuroectoderm of BMP target genes MSX1, ID1 and ID3 in patterned single hES patterning in micropatterned colonies. Together, our data support cells with greater spreading areas (Supplementary Fig. 13). Similar that autonomous regionalization of NE and NPB cells should prob- observations about BMP target genes were also obtained when ably be attributed to cell-fate-determinant, position-dependent seeding hES cells on vitronectin-coated flat PDMS surfaces at dif- information perceived by differentiating cells that dictates the bifur- ferent plating densities to effectively modulate spreading area and cation dynamics of NE and NPB lineage commitments. cytoskeletal contractility (Supplementary Fig. 13).

a ZO-1 p-Myosin Actin Micropost b Control Bleb P = 4.19 ×10–10 P = 0.582 1,600 3,000 ) 2 1,200 ipheral

r 2,000

Pe 800 ncell = 2,201 1,000 ncell = 1,070 n = 773 400 cell n = 198

Cell area ( µ m cell Control 0 0 l l

Centra Centra Peripheral Peripheral Central

c Control Bleb ipheral r Pe

action stress map ncolony = 19 ncolony = 19 Tr Bleb 0–3 (kPa) Central

10 nN

d DAPI PAX6 PAX3 ZIC1 MSX1 Staining image Ble b units ) ry Intensity map ncolony = 30 ncolony = 30 ncolony = 30 ncolony = 37 ncolony = 16 0–1 (arbitra

Fig. 2 | Self-organization of morphogenetic factors controls neuroectoderm patterning. a, Representative fluorescence images showing staining for ZO-1, phosphorylated myosin (p-myosin) and actin, as well as micropost sensors at colony peripheral and central regions at day 4. hPS cells were cultured in neural induction medium supplemented with either DMSO (control) or blebbistatin (Bleb; 10 µM). Circular colonies were divided into two concentric zones ('central' versus 'peripheral') with equal widths. White dashed lines mark the colony periphery. Traction force vectors were superimposed onto individual micropost force sensors. Experiments were repeated three times with similar results. Scale bars, 40 µm (fluorescence micrographs) and 10 µm (micropost images). b, Box-and-whisker plots showing projected cell spreading area at day 4 under different conditions as indicated (box, 25–75%; bar- in-box, median; whiskers: 1% and 99%). The number of colonies analysed was pooled from n = 3 independent experiments. P values were calculated using unpaired, two-sided Student’s t-tests. c, Maps showing the spatial distribution of traction stress quantitated using micropost force sensors at day 4 under culture conditions as indicated (see Methods). The number of colonies analysed was pooled from n = 3 independent experiments. Data plotted were the mean values. Scale bar, 100 µm. d, Representative immunofluorescence micrographs and average intensity maps showing colonies at day 9 stained for Pax6, Pax3, Zic1 and Msx1. DAPI counterstained nuclei. White dashed lines mark the colony periphery. Experiments were repeated three times with similar results. The number of colonies analysed was pooled from n = 3 independent experiments. Data in intensity maps were the mean values. Scale bar, 100 µm.

To further investigate the causal link between cell shape and of Pax3+ NPB cells were evident at the colony centre, whereas NE mechanical force and neuroectoderm patterning, a custom- differentiation was completely inhibited there (Fig. 4e). Together, designed microfluidic cell stretching device was developed and these results unambiguously demonstrate that cell shape and implemented for stretching central regions of micropatterned cell mechanical force can directly activate BMP-SMAD signalling and colonies32 (Fig. 4a and Supplementary Fig. 14). Continuous stretch- thus repress NE but promote NPB differentiation. ing with a square-wave pattern (pulse width of 2 h and period of 4 h) We next modified the neural induction medium to elucidate and a 100% stretch amplitude was applied starting from day 2 under the role of BMP signalling in guiding neuroectoderm patterning. the default neural induction condition (Supplementary Fig. 14). Specifically, modulating LDN concentration in neural induction Indeed, mechanical stretching enhanced cytoskeletal contractility at medium resulted in a dose-dependent BMP signalling response. the colony central region, as reflected by stronger staining of phos- With high-dose LDN (1 µM), BMP-SMAD signalling was com- phorylated myosin and actin filaments, compared to unstretched pletely repressed across entire colonies at day 4 (Supplementary controls (Fig. 4b). Furthermore, stretching of the colony central area Fig. 15). Without LDN supplementation, most cells across entire effectively rescued nuclear p-SMAD 1/5 and thus BMP activity at colonies showed prominent nuclear p-SMAD 1/5 at day 4, consis- the colony central region (Fig. 4c,d). By day 8, a significant number tent with BMP activity-inducing properties of the KnockOut Serum

a Control Bleb Low density Peripheral CentralPeripheral Central Peripheral Central API D p-SMAD 1/5

80 80 60 60 80 80

40 40 30 30 40 40 (a.u.) API intensity

D 0 0 0 0 0 0

80 60 80 60 60 60

50 30 50 30 30 30 p-SMAD 1/5

intensity (a.u.) 20 0 20 0 0 0 010200 10 20 01020 01020 010200 10 20 Distance (μm) Distance (μm) Distance (μm) 4.8 × 10–5 b –6 –4 c d 100 4.9 × 10 4.9 × 10 DAPI p-SMAD 1/5 Merge Nuclear > cytoplasmic 0.065 Nuclear < cytoplasmic Low density n = 3 80 (ncolony = 24) 0.155 n = 3 0.003 n = 3 n = 3 60 0.411 0.021 500 100

Bleb ) ) 40 (ncolony = 24) 2 80 clear p-SMAD 1/ 5 nu ercentage of cells with 20 60

P Control (n = 24) colony 40

0 Cell area ( μ m

0–50 20 ercentage of cells (% P

50–100 1,600 100–150 150–200 0 Distance from colony 500 700 1,100 1,600 centroid ( m) μ 2 Cell area (μm )

Fig. 3 | Mechanics-guided neuroectoderm patterning is mediated by BMP-SMAD signalling. a, Representative immunofluorescence images showing colony central and peripheral zones at day 4 stained for phosphorylated SMAD 1/5 (p-SMAD 1/5). hPS cells were plated at either 20,000 cells cm−2 (control) or 5,000 cells cm−2 (low density) and were cultured in neural induction medium supplemented with either DMSO or blebbistatin (Bleb; 10 µM). Red and yellow rectangles highlight selected peripheral and central regions, respectively, where fluorescence intensities of DAPI and p-SMAD 1/5 were measured along white solid lines drawn across these selected areas. Scale bar, 40 µm. Experiments were repeated three times with similar results. b, Percentage of cells with nuclear p-SMAD 1/5 as a function of distance from colony centroid. Based on distance of nuclei from colony centroid, cells were grouped into four concentric zones with equal widths as indicated. The number of colonies analysed was pooled from n = 3 independent experiments. Data were plotted as the mean ± s.e.m. P values were calculated between Bleb versus control and low density versus control using unpaired, two-sided Student’s t-tests. c, Representative immunofluorescence micrographs showing patterned circular single hPS cells with defined spreading areas (500 µm2 versus 1,600 µm2) stained for p-SMAD 1/5. White dashed lines mark the cell shape. Scale bar, 20 µm. Experiments were repeated three times with similar results. d, Bar plot showing percentages of cells with dominant nuclear or cytoplasmic p-SMAD 1/5 as a function of cell spreading area. n = 3 independent experiments. Data were plotted as the mean ± s.e.m.

Replacement (KSR) medium used in the default neural induction day 9 at colony centre, NE differentiation was completely inhibited, condition33 (Supplementary Fig. 15). Similarly, replacing LDN with whereas NPB differentiation was drastically promoted (Fig. 5c). In BMP4 (25 ng ml−1) in neural induction medium, a condition that distinct contrast, replacing KSR serum in the default neural induc- promotes BMP signalling, induced prominent nuclear accumulation tion condition with the chemically defined Essential 6 medium, of p-SMAD 1/5 across entire colonies (Fig. 5a,b). Consequently, by a condition that does not promote BMP signalling34, inhibited

a b Before stretching Control Stretch Cells Peripheral Central Peripheral Central Vitronectin PDMS membrane Cover slip

Microchamber DAPI During stretching

Δp d/2 d p-Myosin

c Control Stretch Peripheral Central Peripheral Central Actin DAPI

e Control (ncolony = 25) Stretch (ncolony = 22) Staining image Intensity map Staining image Intensity map p-SMAD 1/5 DAPI

d 90 –7 4.2 × 10 0.495 Stretch 0.034 (n = 22) 60 colony 4.9 × 10–4 PAX6

30 Unstretched

nuclear p-SMAD 1/5 (ncolony = 27) Percentage of cells with 0

0–50 PAX3 50–100 100–150150–200 Distance from colony centroid (µm) 0–1 (arbitrary units)

Fig. 4 | Mechanical force is sufficient for activating BMP-SMAD signalling and inducing NPB cell differentiation. a, Schematic of a microfluidic device to apply stretching forces to central zones of micropatterned hPS cell colonies. b, Representative fluorescence micrographs showing central and peripheral zones of unstretched (control) and stretched (stretch) micropatterned colonies at day 4 stained for p-myosin and actin. DAPI counterstained nuclei. Scale bar, 40 µm. Experiments were repeated three times with similar results. c, Representative immunofluorescence images showing central and peripheral zones of unstretched (control) and stretched (stretch) micropatterned colonies at day 4 stained for phosphorylated SMAD 1/5 (p-SMAD 1/5). DAPI counterstained nuclei. Scale bar, 40 µm. Experiments were repeated four times with similar results. d, Percentage of cells with nuclear p-SMAD 1/5 as a function of distance from colony centroid under stretch and unstretched control conditions as indicated. Based on distance of nuclei from colony centroid, cells were grouped into four concentric zones with equal widths as indicated. The number of colonies analysed was pooled from n = 4 independent experiments. Data were plotted as the mean ± s.e.m. P values were calculated between stretch and unstretched control conditions using unpaired, two- sided Student’s t-tests. e, Representative immunofluorescence micrographs and average intensity maps showing unstretched (control) and stretched (stretch) colonies at day 8 stained for Pax6 and Pax3. DAPI counterstained nuclei. White dashed lines mark the colony periphery. Experiments were repeated three times with similar results. The number of colonies analysed was pooled from n = 3 independent experiments. Data in intensity maps were plotted as the mean. Scale bar, 100 µm. nuclear localization of p-SMAD 1/5 at day 4 and promoted NE dif- We further seeded hES cells on flat PDMS surfaces uni- ferentiation at day 9 across entire colonies (Fig. 5a–c). Consistently, formly coated with vitronectin at a low (5,000 cells cm−2) or high NPB differentiation across entire colonies by day 9 was completely (50,000 cells cm−2) density condition. Consistent with our previ- suppressed (Fig. 5c). ous data, the high density condition significantly down-regulated Gene expression analysis further confirmed the effect of expression of NPB markers PAX3 and SOX9 compared with the low replacing KSR with Essential 6 on repressing BMP4 expression density condition (Supplementary Fig. 18). However, exogenously (Supplementary Fig. 16). However, replacing LDN with BMP4 in activating BMP by supplementing BMP4 (25 ng ml−1) effectively the default neural induction medium upregulated both BMP4 and rescued both PAX3 and SOX9 expression (Supplementary Fig. 18). NOGGIN expression (but not BMP2, BMP6 or BMP7), consistent Altogether, our data suggest that morphogenetic cues during emer- with previous reports21,35 (Supplementary Fig. 16). Interestingly, gent neuroectoderm patterning may function upstream of BMP- silencing of BMP4 or NOGGIN expression using siRNA did not SMAD signalling to regulate its transcriptional activation (Fig. 5d). affect neuroectoderm regionalization in micropatterned colonies Exogenous BMP activation can rescue inhibitory effects of confined (Supplementary Fig. 17), excluding endogenous BMP4 or NOGGIN cell shape (or small cell spreading area) and impaired cytoskeletal for instructing NE/NPB patterning. contractile force on BMP signalling.

a b BMP-HI KSR –LDN –KSR (ncolony = 25) –5 + SB + LDN + CHIR + BMP4 + E6 100 1.78 × 10 0.846 (control) (BMP-HI) (BMP-LO)

80 1.39 × 10–5 0.001

DAPI Control

40 (ncolony = 24)

20 nuclear p-Smad 1/5

Percentage of cells with 0.002 –4 5.2 × 10 BMP-LO 0 –5 0.007 2.14 × 10 (ncolony = 21) p-SMAD 1/5

0–50 50–100 100–150 150–200 Distance from colony centroid (μm)

c KSR –LDN –KSR d + SB + LDN + CHIR + BMP4 + E6 (control) (BMP-HI) (BMP-LO) Neural plate border Neuroepithelial cell cell differentiation differentiation (large cell shape and (small cell shape and high contractility) low contractility)

Staining image Actomyosin Stretching PAX3

PAX6 SOX9 SMAD PAX3 SOX10 P P SOX9 SOX10 Confinement p-SMAD Intensity map ncolony = 132 ncolony = 57 ncolony = 24 Zoom-in projected view

Cross-sectional view taining image Colony periphery Colony centre )

PAX3 BMP activity

High Low Intensity ma pS ncolony = 30 ncolony = 57 ncolony = 24 0–1 (arbitrary units

Fig. 5 | BMP-SMAD signalling is required for mechanics-guided neuroectoderm patterning. a, Representative immunofluorescence images showing colonies at day 4 stained for phosphorylated SMAD 1/5 (p-SMAD 1/5). The default neural induction condition (control) was modified as following: − LDN/+BMP4, LDN was replaced with BMP4 (BMP-HI); −KSR/+E6: KSR was replaced with E6 (BMP-LO). CHIR was kept in culture protocols for both BMP-HI and BMP-LO conditions. The zoomed-in image shows a magnified view of the colony central area. Scale bar, 100 µm. Experiments were repeated three times with similar results. b, Percentage of cells with nuclear p-SMAD 1/5 as a function of distance from the colony centroid. The number of colonies analysed was pooled from n = 3 independent experiments. Data represent the mean ± s.e.m. P values were calculated between BMP-HI versus control and BMP-LO versus control using unpaired, two-sided Student’s t-tests. c, Representative fluorescence micrographs and average intensity maps showing colonies at day 9 stained for Pax6 and Pax3. Scale bar, 100 µm. Experiments were repeated three times with similar results. The number of colonies analysed was pooled from n = 3 independent experiments. Data in intensity maps represent the mean values. d, Geometrical confinement leads to self- organization of morphogenetic factors. Increased cell shape and contractile force at colony periphery result in nuclear accumulation and transcriptional activation of p-SMAD 1/5, which in turn up-regulates NPB specifier genes including PAX3, SOX9 and SOX10. Confined cell shape with limited contractile force at the colony centre leads to nuclear exclusion of p-SMAD 1/5 and neuroepithelial differentiation. Modulating cell shape by mechanical stretching or geometrical confinement can thus mediate BMP signalling to regulate neuroectoderm patterning.

Our data, together with others17,18,20,21,35, highlight the dependence to prevent tight junction formation35,36 (Supplementary Fig. 19). of fate patterning in hPS cell colonies on both colony geometry and Such treatment has been shown to allow intercellular diffusion of cell density. Etoc et al. recently report a cell density-dependent apically delivered TGF-β ligands to bind basolateral receptors and mechanism for graded BMP signalling involving subcellular lateral- thus restore downstream SMAD activity in high-density epithelial ization of BMP receptors and diffusion of endogenous NOGGIN35. monolayers35,36. However, under this tight-junction disruption con- To explore whether this mechanism might be involved in our model, dition, proper neuroectoderm patterning was still achieved at day 9 tight-junction integrity was disrupted during neural induction by a (Supplementary Fig. 19), supporting that subcellular lateralization brief calcium depletion followed by incubation with Y-27632 (20 µM), of BMP receptors was unlikely to be responsible for establishing a ROCK (Rho-associated protein kinase) inhibitor, a condition known graded BMP signalling in our model.

Recent studies suggest that cell-density-dependent TGF-β inhi- 5. Gilmour, D., Rembold, M. & Leptin, M. From morphogen to morphogenesis bition in epithelial cells is due to crosstalk with the Hippo pathway37. and back. Nature 541, 311–320 (2017). 6. Schroeder, T. E. Neurulation in Xenopus laevis. An analysis and model The Hippo effector YAP/TAZ is also mechanosensitive and can be based upon light and electron microscopy. J. Embryol. Exp. Morphol. 23, 38,39 regulated by both cell shape and mechanical forces . To examine 427–462 (1970). the role of YAP/TAZ, a small molecular inhibitor cerivastatin (CER), 7. Colas, J.-F. & Schoenwolf, G. C. Towards a cellular and molecular which promotes YAP phosphorylation and thus nuclear exclusion40, understanding of neurulation. Dev. Dynam. 221, 117–145 (2001). was supplemented in the default neural induction medium with 8. Munoz-Sanjuan, I. & Brivanlou, A. H. Neural induction, the default model −1 and embryonic stem cells. Nat. Rev. Neurosci. 3, 271–280 (2002). LDN replaced with BMP4 (25 ng ml ) to promote BMP activity 9. Stern, C. D. Neural induction: old problem, new fndings, yet more questions. (Supplementary Fig. 20). Even though nuclear YAP was suppressed Development 132, 2007–2021 (2005). across entire colonies at day 4, nuclear localization of p-SMAD 1/5 10. Bier, E. & De Robertis, E. M. BMP gradients: a paradigm for morphogen- remained evident and all cells across entire colonies differentiated mediated developmental patterning. Science 348, aaa5838 (2015). into Pax3+ NPB cells at day 9. Thus, YAP nuclear translocation did 11. Vijayraghavan, D. S. & Davidson, L. A. Mechanics of neurulation: From classical to current perspectives on the physical mechanics that shape, fold, not correlate with p-SMAD 1/5 response, and YAP was unlikely to and form the neural tube. Birth Defects Res. 109, 153–168 (2017). directly regulate BMP-SMAD signalling and thus neuroectoderm 12. Tesar, P. J. et al. New cell lines from mouse epiblast share defning features patterning. It remains a future goal to determine how geometrical with human embryonic stem cells. Nature 448, 196–199 (2007). confinement leads to spontaneous self-organization of morphoge- 13. O’Leary, T. et al. Tracking the progression of the human inner cell mass netic factors, how cell shape and cytoskeletal contractility activate during embryonic stem cell derivation. Nat. Biotechnol. 30, 278–282 (2012). 41 14. Nakano, T. et al. Self-formation of optic cups and storable stratifed neural BMP-SMAD signalling , and how BMP and WNT signals converge retina from human ESCs. Cell Stem Cell 10, 771–785 (2012). to regulate NPB specification42,43. 15. Lancaster, M. A. et al. Cerebral organoids model human brain development Our micropatterned neuroectoderm developmental model does and microcephaly. Nature 501, 373–379 (2013). not contain non-neural ectoderm or mesenchymal tissues, which 16. Poh, Y.-C. et al. Generation of organized germ layers from a single mouse embryonic stem cell. Nat. Commun. 5, 4000 (2014). may have important roles in instructing neuroectoderm formation 17. Warmfash, A. et al. A method to recapitulate early embryonic spatial 8–10 in vivo . Nonetheless, there are well documented tissue isola- patterning in human embryonic stem cells. Nat. Methods 11, 847–854 (2014). tion studies using in vivo models that support neural induction as 18. Ma, Z. et al. Self-organizing human cardiac microchambers mediated by an autonomous process44. The origin of NPB cells is also believed geometric confnement. Nat. Commun. 6, 7413 (2015). to arise from neural but not epidermal ectoderm27,29. Altogether, 19. Shao, Y. et al. Self-organized amniogenesis by human pluripotent stem cells in a biomimetic implantation-like niche. Nat. Mater. 16, our neuroectoderm developmental model is useful for studying 419–425 (2017). self-organizing principles involved in autonomous patterning and 20. Shao, Y. et al. A pluripotent stem cell-based model for post-implantation regionalization of neuroectoderm tissues. human amniotic sac development. Nat. Commun. 8, 208 (2017). In this work, we demonstrate that autonomous patterning of 21. Tewary, M. et al. A stepwise model of reaction-difusion and positional- neuroectoderm tissue with proper NP and NPB regionalization information governs self-organized human peri-gastrulation-like patterning. Development 144, 4298–4312 (2017). emerges de novo as the tissue physically takes shape and self-assem- 22. Chambers, S. M. et al. Highly efcient neural conversion of human ES bles in pre-patterned geometrical confinements. Self-organization and iPS cells by dual inhibition of Smad signaling. Nat. Biotechnol. 27, of morphogenetic cues, including cell shape and cytoskeletal con- 275–280 (2009). tractility, can directly feed back to mediate BMP activity and thus 23. Mica, Y. et al. Modeling neural crest induction, melanocyte specifcation, dictate spatial regulation of neuroectoderm patterning. Colony and disease-related pigmentation defects in hESCs and patient-specifc iPSCs. Cell Rep. 3, 1140–1152 (2013). geometry can directly influence cell signalling and cell–cell com- 24. Tchieu, J. et al. A modular platform for diferentiation of human PSCs into all munication through regulatory mechanisms involving dynamic major ectodermal lineages. Cell Stem Cell 21, 399–410 (2017). morphogenetic cues and diffusible signals17,18,20,21,35. Such signalling 25. Hong, C. S. & Saint-Jeannet, J. P. Te activity of Pax3 and Zic1 regulates crosstalk involving both biophysical and biochemical determinants three distinct cell fates at the neural plate border. Mol. Biol. Cell 18, may be important in controlling patterning networks to ensure the 2192–2202 (2007). 26. Sun, Y. et al. Hippo/YAP-mediated rigidity-dependent motor neuron remarkable robustness and precision of tissue self-organization in diferentiation of human pluripotent stem cells. Nat. Mater. 13, vivo1–5. We envision that our neuroectoderm developmental model 599–604 (2014). will facilitate future efforts in generating theoretical frameworks 27. Sauka-Spengler, T. & Bronner-Fraser, M. A gene regulatory network that integrate knowledge at the mechanical, cellular and gene-reg- orchestrates neural crest formation. Nat. Rev. Mol. Cell Biol. 9, 17,18,20,21,35 557–568 (2008). ulatory levels . Future mechanistic investigations of tissue 28. Walther, C. & Gruss, P. PAX-6, a murine paired box gene, is expressed in the mechanics-guided neuroectoderm patterning will therefore help developing CNS. Development 113, 1435–1449 (1991). advance the fundamental understanding of neural development 29. Milet, C. & Monsoro-Burq, A. H. Neural crest induction at the neural plate and disease. border in vertebrates. Dev. Biol. 366, 22–33 (2012). 30. Fu, J. et al. Mechanical regulation of cell function with geometrically Methods modulated elastomeric substrates. Nat. Methods 7, 733–736 (2010). 31. Wilson, P. A. et al. Concentration-dependent patterning of the Xenopus Methods, including statements of data availability and any asso- ectoderm by BMP4 and its signal transducer Smad1. Development 124, ciated accession codes and references, are available at https://doi. 3177–3184 (1997). org/10.1038/s41563-018-0082-9. 32. Michielin, F. et al. Microfuidic-assisted cyclic mechanical stimulation afects cellular membrane integrity in a human muscular dystrophy in vitro model. Received: 8 May 2017; Accepted: 16 April 2018; RSC Adv. 5, 98429–98439 (2015). 33. Xu, R.-H. et al. Basic FGF and suppression of BMP signaling sustain Published online: 21 May 2018 undiferentiated proliferation of human ES cells. Nat. Methods 2, 185–190 (2005). References 34. Lippmann, E. S., Estevez-Silva, M. C. & Ashton, R. S. Defned human 1. Wozniak, M. A. & Chen, C. S. Mechanotransduction in development: a pluripotent stem cell culture enables highly efcient neuroepithelium growing role for contractility. Nat. Rev. Mol. Cell Biol. 10, 34–43 (2009). derivation without small molecule inhibitors. Stem Cells 32, 2. Keller, R. Physical biology returns to morphogenesis. Science 338, 1032–1042 (2014). 201–203 (2012). 35. Etoc, F. et al. A balance between secreted inhibitors and edge sensing controls 3. Heller, E. & Fuchs, E. Tissue patterning and cellular mechanics. J. Cell Biol. gastruloid self-organization. Dev. Cell 39, 302–315 (2016). 211, 219–231 (2015). 36. Nallet-Staub, F. et al. Cell density sensing alters TGF-β signaling in a 4. Chan, C. J., Heisenberg, C.-P. & Hiiragi, T. Coordination of morphogenesis cell-type-specifc manner, independent from Hippo pathway activation. and cell-fate specifcation in development. Curr. Biol. 27, R1024–R1035 (2017). Dev. Cell 32, 640–651 (2015).

37. Varelas, X. et al. Te Crumbs complex couples cell density sensing to and the Department of Mechanical Engineering at the University of Michigan. The Hippo-dependent control of the TGF-β-SMAD pathway. Dev. Cell 19, Lurie Nanofabrication Facility at the University of Michigan, a member of the National 831–844 (2010). Nanotechnology Infrastructure Network funded by the National Science Foundation, 38. Dupont, S. et al. Role of YAP/TAZ in mechanotransduction. Nature 474, is acknowledged for support in microfabrication. 179–183 (2011). 39. Wada, K.-I. et al. Hippo pathway regulation by cell morphology and stress fbers. Development 138, 3907–3914 (2011). Author contributions 40. Sorrentino, G. et al. Metabolic control of YAP and TAZ by the mevalonate Y. Sun, X.X. and J.F. designed experiments; X.X. and Y. Sun performed differentiation pathway. Nat. Cell Biol. 16, 357–366 (2014). assays; A.R.-I. and Y. Sun developed MATLAB scripts for image processing; X.X., Y. Sun, 41. Kopf, J. et al. BMP growth factor signaling in a biomechanical context. K.M.A.Y., Y.Z., S.W. and Y. Shao generated and analysed gene expression data; X.X. and BioFactors 40, 171–187 (2014). Y.Y. conducted cell migration assays; L.S. provided Sox10-EGFP cells; Y. Sun, X.X., Y.C. 42. de Croz‚, N., Maczkowiak, F. & Monsoro-Burq, A. H. Reiterative AP2a and J.F. analysed data and wrote the manuscript. J.F. supervised the entire project. activity controls sequential steps in the neural crest gene regulatory network. All authors edited and approved the manuscript. Proc. Natl Acad. USA 108, 155–160 (2011). 43. Steventon, B. et al. Diferential requirements of BMP and Wnt signalling during gastrulation and neurulation defne two steps in neural crest Competing interests induction. Development 136, 771–779 (2009). The authors declare no competing interests. 44. Moury, J. D. & Schoenwolf, G. C. Cooperative model of epithelial shaping and bending during avian neurulation: autonomous movements of the neural Additional information plate, autonomous movements of the epidermis, and interactions in the Supplementary information is available for this paper at https://doi.org/10.1038/ neural plate/epidermis transition zone. Dev. Dynam. 204, 323–337 (1995). s41563-018-0082-9. Acknowledgements Reprints and permissions information is available at www.nature.com/reprints. We thank A. Liu for comments on the manuscript. This work is supported in part Correspondence and requests for materials should be addressed to Y.S. or J.F. by the National Science Foundation (CMMI 1129611 and CBET 1149401 to J.F. and Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in CMMI 1662835 to Y. Sun), the American Heart Association (12SDG12180025 to J.F.) published maps and institutional affiliations.

Methods microfluidic cell stretching device to transfer vitronectin from the stamp to the Culture medium. PluriQ Human Cell Conditioned Medium (MTI-GlobalStem) PDMS membrane on top of the pressurization compartments. was used to support feeder-free growth of hPS cells45. Te growth medium comprised DMEM/F12 (GIBCO), 20% KnockOut Serum Replacement (KSR; Stencil micropatterning. The stencil was generated by punching through-holes GIBCO), β-mercaptoethanol (0.1 mM; GIBCO), glutamax (2 mM; GIBCO), 1% into a PDMS membrane using a 500 µm biopsy puncher (Fisher Scientific). To non-essential amino acids (GIBCO), and human recombinant basic fbroblast generate the PDMS membrane, PDMS prepolymer was spun on a silicon wafer growth factor (bFGF, 4 ng ml−1; GlobalStem) as described previously26. Te neural before PDMS was thermally cured at 60 °C for at least 24 h. The PDMS membrane induction medium comprised growth medium, TGF-β inhibitor SB 431542 (10 µM; was then peeled off the silicon wafer before punching through-holes into the 1 Cayman Chemical) and BMP4 inhibitor LDN 193189 (500 nM; Selleckchem)22. At membrane. Coverslips were immersed in a vitronectin solution (20 µg ml− ) for 1 h day 3, CHIR99021 (3 µM; Cayman Chemical) was added to the neural induction before the PDMS stencil was placed onto the coverslip for cell seeding. medium and was withdrawn at day 423,24. Te motor neuron diferentiation medium 26 comprised N2B27 medium, retinoic acid (RA, 1 µM; Stemcell Technologies) and Immunocytochemistry. As described previously , 4% paraformaldehyde smoothened agonist (SAG, 500 nM; Stemcell Technologies)26. N2B27 medium (Electron Microscopy Sciences) was used for cell fixation before permeabilization comprised 1:1 mixture of DMEM/F12 and neurobasal medium (GIBCO), 1% N2 with 0.1% Triton X-100 (Roche Applied Science). Primary antibodies listed in supplement (GIBCO), 2% B-27 supplement (GIBCO), 2 mM glutamax and 1% Supplementary Table 2 were then used for protein detection. For immunolabelling, non-essential amino acids. Te neural crest cell diferentiation medium comprised goat-anti mouse AlexaFluor 488 and/or goat-anti rabbit AlexaFluor 546 secondary N2B27 medium and 3 µM CHIR9902123,24. Culture medium was pre-equilibrated at antibodies were used. To label actin microfilaments, AlexaFluor 555 conjugated 37 °C and 5% CO2 before use. phalloidin (Invitrogen) was used. Cells were stained with 4,6-diamidino-2- phenylindole (DAPI; Invitrogen) to visualize the nucleus. Cell culture. Both H1 hES cell line (WA01, WiCell; NIH registration number: 0043) and Sox10-EGFP bacterial artificial chromosome hES cell reporter line Image analysis. Fluorescence images were recorded using either an inverted (H9; WA09, WiCell; NIH registration number: 0062) were cultured on mitotically epifluorescence microscope (Zeiss Axio Observer Z1; Carl Zeiss MicroImaging) inactive mouse embryonic fibroblasts (MEFs; GlobalStem) in growth medium. equipped with a monochrome charge-coupled device (CCD) camera or an Another H9 hES cell line was cultured using mTeSR1 medium (Stemcell Olympus DSUIX81 spinning disc confocal microscope equipped with an EMCCD Technologies) and lactate dehydrogenase-elevating virus (LDEV)-free hES cell- camera (iXon X3, Andor). Fluorescence images of micropatterned cell colonies qualified reduced growth factor basement membrane matrix Geltrex (Thermo were first cropped using a custom-developed MATLAB program (MathWorks; Fisher Scientific) as described previously19. The hiPS cell line, a gift from P. H. https://www.mathworks.com/) to a uniform circular size (as defined by colony Krebsbach, was cultured in PluriQ Human Cell Conditioned Medium on tissue size) with pattern centroids aligned. Owing to intrinsic inhomogeneous cell culture dishes coated with poly[2-(methacryloyloxy)ethyl dimethyl-(3-sulfopropyl) seeding, multilayered cellular structures would inevitably appear in micropatterned ammonium hydroxide]45. cell colonies. These multilayered colonies were excluded from data analyses. The STEMPRO EZPassage Disposable Stem Cell Passaging Tool (Invitrogen) Fluorescence intensity of each pixel in cropped images was normalized by the was used for passaging the H1 and Sox10-EGFP hES cell lines every 5 d maximum intensity identified in each image. These normalized images were as described previously26. A modified pasteur pipette was used to remove stacked together to obtain average intensity maps. To normalize fluorescence differentiated cells under a stereomicroscope (Olympus). After brief treatment intensities of cell lineage markers by DAPI intensity, the intensity of cell lineage with TrypLE Select (Invitrogen) to release MEFs, the remaining cells were collected markers for each pixel was divided by the corresponding DAPI intensity. These using a cell scraper (BD Biosciences) before being transferred onto a tissue culture DAPI-normalized images were stacked together to obtain average DAPI- dish coated with gelatin (Sigma) and incubated for 45 min. Contaminating MEFs normalized intensity maps. To plot average intensity as a function of distance from would attach to the dish. hES cells in suspension were then collected, centrifuged colony centroid, average intensity maps for circular cell colonies were divided and re-dispersed in growth medium supplemented with Y27632 (10 µM; Enzo Life into 100 concentric zones with equal widths. The average pixel intensity in each Sciences) before cell seeding. For H9 hES and hiPS cells, cells were digested using concentric zone was calculated and plotted against the mean distance of the TrypLE Select before cell seeding19. concentric zone from colony centroid. From average intensity plots of cell lineage The Human Pluripotent Stem Cell Research Oversight Committee at the markers, values of the FWHM were determined as the difference between the two University of Michigan have approved all protocols for the use of hPS cells. All cell radial positions at which the average fluorescence intensity of individual markers lines were authenticated by Cell Line Genetics as karyotypically normal and were is equal to half of its maximum value. The FWHM for average DAPI intensity plots tested negative for mycoplasma contamination using the LookOut Mycoplasma was determined by first calculating the half width at half maximum (HWHM), PCR Detection Kit (Sigma-Aldrich). which was determined as the radial position at which the average DAPI intensity is equal to half of its maximum value at the colony centroid. the FWHM for average Microcontact printing. Soft lithography was used to generate patterned PDMS DAPI intensity plots was then calculated as double the HWHM. stamps from silicon moulds fabricated using photolithography and deep reactive- Colony thickness and nucleus orientation were determined manually with ion etching (DRIE), as described previously30. These PDMS stamps were used to ImageJ (https://imagej.nih.gov/) using confocal images showing X–Z sections generate micropatterned cell colonies and patterned single cells using microcontact of cell colonies stained for N-cadherin. Normalized nucleus dimension was printing. Briefly, to generate patterned cell colonies and single cells on flat PDMS further calculated as the ratio of the height to the width of a circumscribed surfaces, round glass coverslips with a diameter of 18 mm (Fisher Scientific) rectangle bounding the nucleus. Projected cell areas were quantified with the were spin-coated (Spin Coater; Laurell Technologies) with a thin layer of PDMS CellProfiler program in ImageJ using immunofluorescence images showing ZO-1 prepolymer comprising the PDMS base monomer and curing agent (10:1 w/w; staining. Briefly, nuclei stained by DAPI were used for identification of each Sylgard 184, Dow-Corning). PDMS was then thermally cured at 110 °C for at least cell as primary objects in CellProfiler. ZO-1 staining images were first filtered 24 h. In parallel, PDMS stamps were immersed in a vitronectin solution (20 µg ml−1; to remove background before being segregated using a propagation method to Trevigen) for 1 h before being blown dry under nitrogen. Vitronectin has been determine cell area. To determine projected cell area for central and peripheral reported to support self-renewal of hPS cells46. Vitronectin-coated PDMS stamps zones of cell colonies, a binary circular mask with a diameter of half of the actual were then placed in conformal contact with ultraviolet ozone-treated PDMS on pattern diameter was used to crop the original cell colony images. Cells falling coverslips (ozone cleaner; Jelight). After removing PDMS stamps, coverslips were on the mask boundary were excluded from quantification. Immunofluorescence sterilized using ethanol (Fisher Scientific). Protein adsorption to PDMS surfaces images showing p-SMAD 1/5 staining were analysed manually using ImageJ. Cells not coated with vitronectin was prevented by immersing coverslips in 0.2% with nuclear p-SMAD 1/5 were identified as those showing dominant nuclear Pluronics F127 NF solution (BASF) for 30 min. Coverslips were rinsed with PBS fluorescence and absence of cytoplasmic fluorescence. Conversely, cells with before placed into tissue culture plates for cell seeding. For micropatterned cell cytoplasmic p-SMAD 1/5 were identified as those showing absence of nuclear colonies, PDMS stamps containing circular patterns with diameters of 300, 400, fluorescence. To quantify the spatial distribution of BMP-SMAD activation, 500 and 800 µm were used. For patterned single cells, PDMS stamps containing circular colonies were divided into four concentric zones with equal widths. The circular patterns with diameters of 25, 30, 37 and 45 µm were used. percentage of cells with nuclear p-SMAD 1/5 in each zone was then calculated and To apply continuous stretching to the central zone of micropatterned cell plotted against the distance from colony centroid. colonies, microcontact printing was performed to print circular adhesive patterns with a diameter of 400 µm onto the deformable PDMS membrane on Tracking cell migration and unbiased random walk model. On day 0 before cell top of pressurization compartments (with a diameter of 200 µm) in a custom- seeding, hPS cells were labelled with CellTracker Red CMTPX Dye (ThermoFisher designed microfluidic cell stretching device. To this end, a custom desktop aligner Scientific) for 30 min. Labelled hPS cells were mixed with unlabelled cells at a ratio designed for fabrication of multilayer microfluidic devices was used47. Briefly, the of 1:7 before the cell mixture was seeded onto coverslips containing micropatterned vitronectin-coated PDMS stamp and the microfluidic cell stretching device were circular adhesive islands. From day 2 (t = 0 h) to day 4, live cell imaging was mounted onto the top and bottom layer holders of the aligner, respectively. Under conducted using the Zeiss Axio Observer Z1 inverted epifluorescence microscope a digital microscope, the X/Y/θ stage holding the bottom layer holder was carefully enclosed in the XL S1 incubator (Carl Zeiss MicroImaging) to maintain cell adjusted to align the PDMS stamp and the microfluidic cell stretching device. culture at 37 °C and 5% CO2. Both bright field and fluorescence images were The PDMS stamp was then gently pressed to achieve conformal contact with the recorded every 20 min for a total of 41 h. All time-lapse images were reconstructed

Nature Materials | www.nature.com/naturematerials © 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved. NATure MATerIAlS Articles using Image-Pro Plus (http://www.mediacy.com/imageproplus) to generate TIFF being treated with air plasma (Plasma Prep II; SPI Supplies) and bonding together. stacks for cell migration tracking. For each image, peripheries of each individual In parallel, another PDMS block was prepared, and an inlet for fluid connection CellTracker labelled cells were marked manually. Positions of cell centroids were was punched into the PDMS block with a 0.5 mm biopsy punch. After treating then identified from each image to calculate end-to-end cell displacement D both with air plasma, the PDMS block and the PDMS structural layer were bonded and radial displacement during cell migration as a function of time. For radial together with their fluid inlets aligned manually. The microfluidic cell stretching displacement calculations, cells were divided into two groups based on their radial device was baked at 110 °C for at least another 24 h to ensure robust bonding positions relative to the colony centroid at t = 0 h (R0) (‘central’ 0 ≤ R0 ≤ 100 μm; between different layers. Deionized water was injected into the microfluidic cell ‘peripheral’ 100 < R0 ≤ 200 μm). If labelled hPS cells divided during cell migration stretching device before applying pressure through a microfluidic pressure pump tracking, one of the daughter cells was randomly selected to continue tracking. (AF1, Elveflow). Elveflow Smart Interface software (https://www.elveflow.com/) An unbiased random walk model was used for modelling cell migration was used for programming the pressure pump for continuous cell stretching with a during emergent neuroectoderm patterning in micropatterned circular hPS cell square-wave pattern (pulse width of 2 h, period of 4 h and 50% duty cycle). colonies. In this model, cells move randomly without any preferred migration 26 direction. Cell centroids at each time step n (denoted here as Dn) can be calculated RNA isolation and qRT-PCR analysis. As described previously , RNA was as Dnn=+D −1 AV γ, where A is a random variable following a uniform distribution extracted from cells using the RNeasy kit (Qiagen) following the manufacturer’s on [0, 1] and γ is a random unit direction vector. Parameter V denotes the instructions. A CFX Connect SYBR Green PCR Master Mix system (Bio-Rad) maximum cell centroid displacement during each time step, and it can be obtained was used for qRT-PCR. For relative quantification, human TBP primer was used through least-squares fitting of the experimental data of mean square end-to-end as an endogenous control. An arbitrary Ct value of 40 was assigned to samples in 22 cell displacement D2 using the expression D =∕Vt 4. which no expression was detected. Relative expression levels were determined by calculating 2−∆∆Ct with the corresponding s.e.m. All analyses were performed with Cell viability and proliferation assays. Cell viability was determined using the at least three biological replicates and three technical replicates. Live/Dead, Viability/Cytotoxicity Kit (ThermoFisher Scientific) according to the manufacturer’s instructions. On day 4 and day 8, culture medium was replaced Western blotting. As described previously26, whole cell lysates were prepared with fresh medium containing ethidium homodimer (EthD-1; 4 µM). After and separated on SDS-polyacrylamide gel before being transferred to PVDF incubation for 30 min, fluorescence images were recorded using the Zeiss Axio membranes. PVDF membranes were then incubated with blocking buffer (Li-Cor) Observer Z1 inverted epifluorescence microscope to quantify the number of dead for 1 h before being soaked with primary antibodies overnight at 4 °C. Blots were cells in each colony. then incubated with IRDye secondary antibodies (Li-Cor) for 1 h. An Odyssey The Click-iT EdU Alexa Fluor 488 Imaging Kit (ThermoFisher Scientific) was Sa Infrared Imaging System (Li-Cor) was used for protein expression detection. used to measure cell proliferation according to the manufacturers’ instructions. Uncropped scans of Western blots are shown in Supplementary Fig. 2. Briefly, on day 3, half of the culture medium was replaced with fresh medium containing EdU (20 µM). Cells were incubated for 2 h before being fixed, siRNA knockdown. As described previously26, hPS cells were transfected with permeabilized and incubated with the Click-iT reaction cocktail for 30 min. Cell BMP4 or NOGGIN siRNA and scramble control using Viromer Blue (OriGene). nuclei were counterstained with DAPI. Cells were then examined under fluorescence Briefly, cells were plated at 80% confluence on vitronectin-coated 6-well plates microscopy (Zeiss Axio Observer Z1) to detect EdU-stained cell nuclei. and subjected to siRNA transfection the next day. After 24 h, transfected cells were seeded onto coverslips containing micropatterned circular adhesive islands. Three Traction force measurement. Traction force was analysed using PDMS micropost additional siRNA treatments were conducted at days 2, 4 and 6 after cell seeding. arrays (PMAs) as described previously30. Briefly, PMAs were first prepared for cell attachment using microcontact printing to coat micropost top surfaces with Pharmacological treatment. To inhibit actomyosin contractility, blebbistatin (10 µ vitronectin. To quantify the traction forces of micropatterned cell colonies or M in DMSO; Cayman Chemical) was added to culture medium from day 1 till single cells, microcontact printing was used to define circular adhesive patterns cell fixation. To modulate BMP signalling, recombinant human BMP4 (25 ng ml−1; of different sizes. PMAs were labelled with Δ9-DiI (5 μg ml−1; Invitrogen) for 1 h. R&D Systems) was added to the neural induction medium. Protein adsorption to all PDMS surfaces not coated with vitronectin was prevented by incubating in 0.2% Pluronics F127 NF solution (BASF) for 30 min. Statistics. Statistical analysis was performed using Excel (Microsoft; https://www. Live cell imaging was conducted using the Zeiss Axio Observer Z1 inverted microsoft.com). The statistical significance between two groups was analysed by epifluorescence microscope enclosed in the XL S1 incubator. Images of micropost a two-sided Student's t-test. In all cases, a P value of less than 0.05 was considered tops were recorded using a 40× (for single cells) or 20× (for colonies) objective statistically significant. (Carl Zeiss MicroImaging). All images were recorded at day 4 and were analysed using a custom-developed MATLAB program (MathWorks), as described Reporting Summary. Further information on experimental design is available in previously48, to obtain traction force maps. To determine average traction stress the Nature Research Reporting Summary linked to this article. maps for micropatterned cell colonies, individual traction force maps were first adjusted to the same size before being stacked together to obtain average traction Code availability. MATLAB scripts used in this work are available from the stress maps. To plot average traction stress as a function of distance from the corresponding authors upon reasonable request. colony centroid, circular colonies were first divided into 20 concentric zones with equal widths. The average traction stress in each concentric zone was then Data availability. Data supporting the findings of this study are available within calculated and plotted against the mean distance of the concentric zone from the the article and its Supplementary Information files and from the corresponding colony centroid. authors upon reasonable request.

Microfluidic cell stretching device. The microfluidic cell stretching device comprised a PDMS structural layer, a PDMS inlet block and a glass coverslip. References The PDMS structural layer, which contained a microfluidic network for applying 45. Villa-Diaz, L. G. et al. Synthetic polymer coatings for long-term growth of pressures to simultaneously activate 64 pressurization compartments to induce human embryonic stem cells. Nat. Biotechnol. 28, 581–583 (2010). PDMS membrane deformation, was fabricated using soft lithography. Briefly, 46. Braam, S. R. et al. Recombinant vitronectin is a functionally defned substrate the PDMS prepolymer was spin-coated onto a silicon mould generated using that supports human embryonic stem cell self-renewal via αVβ5 integrin. photolithography and DRIE. The PDMS layer was thermally cured at 110 °C for at Stem Cells 26, 2257–2265 (2008). least 24 h before being peeled off the silicon mould. An inlet for fluid connections 47. Li, X. et al. Desktop aligner for fabrication of multilayer microfuidic devices. was then punched into the PDMS structural layer using a 1 mm biopsy punch Rev. Sci. Instrum. 86, 075008 (2015). (Fisher Scientific). Both the coverslip and the PDMS structural layer were briefly 48. Weng, S. et al. Mechanosensitive subcellular rheostasis drives emergent cleaned with 100% ethanol (Fisher Scientific) and blown dry under nitrogen before single-cell mechanical homeostasis. Nat. Mater. 15, 961–967 (2016).

Corresponding author(s): Jianping Fu and Yubing Sun

Life Sciences Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form is intended for publication with all accepted life science papers and provides structure for consistency and transparency in reporting. Every life science submission will use this form; some list items might not apply to an individual manuscript, but all fields must be completed for clarity. For further information on the points included in this form, see Reporting Life Sciences Research. For further information on Nature Research policies, including our data availability policy, see Authors & Referees and the Editorial Policy Checklist.

Please do not complete any field with "not applicable" or n/a. Refer to the help text for what text to use if an item is not relevant to your study. For final submission: please carefully check your responses for accuracy; you will not be able to make changes later.

` Experimental design 1. Sample size Describe how sample size was determined. All experiments were conducted with at least two independent experiments and multiple biological replicates. Sample sizes were determined based on our previous experience and similar studies of other groups. Sample sizes were determined as sufficient since they led to similar results. 2. Data exclusions Describe any data exclusions. Due to intrinsic inhomogeneous cell seeding, some multilayered cellular structures would inevitably appear in micropatterned cell colonies. These multilayered colonies were excluded from data analyses. When quantifying cell morphological features and nuclear staining as a function of different regions of micropatterned cell colonies, those cells falling on region boundaries were excluded from quantification. This information is explained in the "Image analysis" section of the Methods. 3. Replication Describe the measures taken to verify the reproducibility Reported results were repeated and confirmed for at least two independent experiments. of the experimental findings. Key experimental findings were reliably reproduced by two investigators involved in this work. 4. Randomization Describe how samples/organisms/participants were Samples were randomly allocated to control and different experimental groups (see allocated into experimental groups. Methods). However, no particular randomization method was used in this work. 5. Blinding Describe whether the investigators were blinded to Investigators were not blinded to group allocation, as no animal/human studies were group allocation during data collection and/or analysis. conducted in this manuscript. Note: all in vivo studies must report how sample size was determined and whether blinding and randomization were used. November 2017

1 6. Statistical parameters nature research | life sciences reporting summary For all figures and tables that use statistical methods, confirm that the following items are present in relevant figure legends (or in the Methods section if additional space is needed). n/a Confirmed

The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement (animals, litters, cultures, etc.) A description of how samples were collected, noting whether measurements were taken from distinct samples or whether the same sample was measured repeatedly A statement indicating how many times each experiment was replicated The statistical test(s) used and whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section. A description of any assumptions or corrections, such as an adjustment for multiple comparisons Test values indicating whether an effect is present Provide confidence intervals or give results of significance tests (e.g. P values) as exact values whenever appropriate and with effect sizes noted. A clear description of statistics including central tendency (e.g. median, mean) and variation (e.g. standard deviation, interquartile range) Clearly defined error bars in all relevant figure captions (with explicit mention of central tendency and variation)

See the web collection on statistics for biologists for further resources and guidance.

` Software Policy information about availability of computer code 7. Software Describe the software used to analyze the data in this MATLAB, ImageJ, Image-Pro Plus were used for image analysis. Elveflow Smart Interface was study. used for programming pressure pump. Excel (Microsoft) was used for statistics.

For manuscripts utilizing custom algorithms or software that are central to the paper but not yet described in the published literature, software must be made available to editors and reviewers upon request. We strongly encourage code deposition in a community repository (e.g. GitHub). Nature Methods guidance for providing algorithms and software for publication provides further information on this topic.

` Materials and reagents Policy information about availability of materials 8. Materials availability Indicate whether there are restrictions on availability of All materials used in this work are commercially available. There is no restriction on unique materials or if these materials are only available availability of any materials used in this work. for distribution by a third party. 9. Antibodies Describe the antibodies used and how they were validated Only certified and company-validated antibodies were used in this work for Western blotting for use in the system under study (i.e. assay and species). (WB) or immunocytochemistry (ICC): ZO-1, Invitrogen (33-9100), 1:50 (ICC) PAX6, Abcam (ab78545), 1:200 (ICC) PAX3, Novus (NBP1-32944), 1:200 (ICC) ZIC1, Novus (NB600-488), 1:200 (ICC) MSX1, Novus (NBP2-30052), 1:200 (ICC) p-SMAD 1/5, Millipore (AB3848), 1:500 (WB) and 1:100 (ICC) p-MLC, Cell Signaling (3671S), 1:50 (ICC) GAPDH, Santa-Cruz Biotechnology (sc-25778), 1:200 (WB) N-CADHERIN, Abcam (ab12221), 1:100 (ICC) and 1:500 (WB) E-CADHERIN, BD Biosciences (610181), 1:100 (ICC) and 1:500 (WB) OLIG2, Santa-Cruz Biotechnology (sc-48817), 1:100 (ICC) NESTIN, Santa-Cruz Biotechnology (sc-23927), 1:100 (ICC) MAP2, Sigma-Aldrich (M1406-.2ML), 1:100 (ICC)

SOX10, Cell Signaling (89356S), 1:100 (ICC) November 2017 AP2α, Santa-Cruz Biotechnology (sc-12726), 1:100 (ICC)

The antibody information (including species, application, and catalog number) has been provided in Supplementary Information (Supplementary Table 2). All antibodies have been validated by the companies from which they were purchased from. The subcellular localization of all the proteins analyzed in this work has been previously reported. This information was used to further validate the specificity of antibodies.

2 10. Eukaryotic cell lines nature research | life sciences reporting summary a. State the source of each eukaryotic cell line used. The following cell lines were used in this work: H1 hES cell line (WA01, WiCell; NIH registration number: 0043) H9 hES cell line (WA09, WiCell; NIH registration number: 0062) A hiPS cell line that is reported in Villa-Diaz, L. G. et al. Nat. Biotechnol. 28, 581 (2010)

b. Describe the method of cell line authentication used. All hPS cell lines have been authenticated by the original sources and also authenticated in- house by immunostaining for pluripotency markers and successful differentiation to the three germ layer cells.

c. Report whether the cell lines were tested for All cell lines used in this work have been tested negative for mycoplasma contamination. mycoplasma contamination. d. If any of the cell lines used are listed in the database No commonly misidentified cell lines listed by ICLAC were used in this work. of commonly misidentified cell lines maintained by ICLAC, provide a scientific rationale for their use.

` Animals and human research participants Policy information about studies involving animals; when reporting animal research, follow the ARRIVE guidelines 11. Description of research animals Provide all relevant details on animals and/or Irrelevant to this work. This work doesn't involve animal or animal-derived materials. animal-derived materials used in the study.

Policy information about studies involving human research participants 12. Description of human research participants Describe the covariate-relevant population Irrelevant to this work. This work doesn't involve human or human samples. characteristics of the human research participants. November 2017